sign in sign up
<<
Home , Philosophy

The of

Richard Feldman

Philosophy and Phenomenological , Vol. 60, No. 3. (May, 2000), pp. 667-695.

Stable URL: http://links.jstor.org/sici?sici=0031-8205%28200005%2960%3A3%3C667%3ATEOB%3E2.0.CO%3B2-H

Philosophy and Phenomenological Research is currently published by International Phenomenological .

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact may be obtained at http://www.jstor.org/journals/ips.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly from around the . The Archive is supported by libraries, scholarly , publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in . For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Tue Aug 28 10:01:58 2007 Philosophy and Pltenoi~teitologicalResearclt Vol. LX, No. 3, May 2000

The Ethics of Belief

RICHARD FELDMAN of Rochester

In this paper I address a few of the many questions that fall under the general heading of "the ethics of belief." In section I I will discuss the adequacy of what has come to be known as the "deontological conception of epistemic justification" in the light of our apparent lack of voluntary control over what we believe. In section I1 I'll defend an evidentialist view about what we ought to believe. And in section I11 I will briefly discuss apparent conflicts between epistemic considerations and moral or other considerations.

I. Epistemic Deontologism' A. The Problenz Our talk about epistemic matters parallels our talk about ethical matters in noteworthy ways. In the case of ethics, we say that a ought to perform a certain , that someone should not do a certain thing, that people have to act in some ways, that they have and , and that they deserve praise or blatne for what they have done. We make seemingly analogous epistemic judgments about, beliefs. We say that a typical well- informed contemporary American ought to believe that the Earth revolves around the Sun and should not believe that the Earth is flat. We say that a person has a right to believe one thing and perhaps a to refrain from believing something else. We sometimes praise those who believe the things they should and we criticize those who fail in their believings. We can describe all these judgments as deontological judgments about beliefs. In "The Deontological Conception of Epistemic Justification," William Alston says that the most natural way to understand epistemic justification is deontol~gical.~By this he means that it is to be understood in terms of epis-

' This section is a shortened version of my paper, "Doxastic Voluntarism and the Ethics of Belief," forthcoming in a collection of papers on epistemic responsibility edited by Matthias Steup. William Alston, "The Deontological Conception of Epistemic Justification," P/lilosophiccrl Pers11ectii~es2 (1988): 257-99. See especially p. 258. Reprinted in William Alston, Epi.sternic Just~ccltion(New York: Cornell University Press, 1989).

SYMPOSIUM 667 temic ", permission, requirement, blame, and the like."3 He regards "requirement,prolzibition, and permission as the basic deontological terms" and the other terms as derivative ones.%lston eventually argues that deonto- logical conceptions are in the end ill-suited to epistemic p~rposes.~Recently, Alvin Plantinga and have independently argued that the viability of a deontological conception of epistemic justification is crucial to the debate between internalists and externalists about epistemic justifi~ation.~ Goldman thinks that a central, but mistaken, line of support for internalist begins with the assumption of a deontological account of justification. Plantinga also argues that internalism derives much of its support from a deontological view of justification.' The merits of the deonto- logical view of epistemic justification thus is of considerable epistemological significance to contemporary . A central problem that both Plantinga and Alston find with deontological judgments about beliefs is that they presuppose that we have voluntary control over what we believe. Yet, reflection on our mental lives suggests that we have no such control. Alston writes of one deontological :

...this conception of epistemic justification is viable only if beliefs are sufficiently under volun- tary control to render such as requirement, permission, obligation, re11roclch, and blame applicable to them. Ry the honored that "Ought implies can", one can be obliged to do A only if one has an effective choice as to whether to do A.'

He goes on to argue that we don't have an effective choice over what we believe. In the process of objecting to Chisholm's views about justification, Plantinga says of a particular that "...whether or not I accept it is simply not up to me; but then accepting this proposition cannot be a way in which I can fulfill my obligation to the , or, indeed, any obligation.. ."' Thus, according to Plantinga, our lack of control over beliefs implies that

"The Deontological Conception of Epistemic Justification", p. 257. Ibid. The most prominent advocate of this sort of analysis is . He defends this view as recently as in his in 1991 paper, "Firth and the Ethics of Relief," Philosophy and Phenornenologiccll Reseclrch 50 (1991): 119-28. For a thorough list of Chisholm's publications on this topic, see Susan Haack, "'The Ethics of Relief' Reconsidered," in Lewis E. Hahn, (ed.) The Philosophy of Roderick Chisholrn (LaSalle: Open Court, 1997), pp. 129-44. See especially footnote 2. See also , "Justified Belief and Epistemically Responsible Action," Philosophicell Review 92 (1983): 33-48 for an influential discussion of the that a belief is epistemically justified provided it is formed in an epistemically responsible way. See Alvin Plantinga, Wctrront: The Current Debate (New York: , 1993) and Alvin Goldman "Internalism Exposed," the Journal of Philosoplzy 96 (1999): 271-93. Wurrctnt: The Current Debate, Chapter 1. "The Deontological Conception of Epistemic Justification," p. 259. Worrctnt: The Cul.rent Debate, p. 38.

668 RICHARD FELDMAN they are not the sort of thing that can be a of obligation, and this undermines Chisholm's deontological conception of epistemic justification. Matthias Steup presents a similar , though he goes on to defend the deontological conception on the grounds that belief is v~luntary.'~' For the purposes of the discussion that follows it will be helpful to distinguish two elements of the just presented. Their target is a deontological conception of justification, a conception according to which epistemic justification is to be understood or analyzed in tarns of the deonto- logical concepts of obligation, requirement, and the like. This conception is "viable", in Alston's terms, only if belief is sufficiently under our voluntary control. Presumably, the deontological conception is viable only if it can be true that we are required to believe things, that we ought not believe other things, and so on. That is, the deontological conception of epistemic justification is viable only if deontological judgments about beliefs are some- true. My goal in what follows is to argue that deontological judgments are sometimes true and that can be made of the idea that we can have epistemic requirements, obligations, and the like. I will not be arguing for the further claim that epistemic justification should be analyzed in deonto- logical terms. We can formulate the key part of the argument at issue in the following way:

The Voluntarism Argument

1. If deontological judgments about beliefs are true, then people have voluntary control over their beliefs.

2. People do not have voluntary control over their beliefs.

3. Deontological judgments about beliefs are not true.

Epistemologists have three kinds of response to this argument open to them: i) they can argue that we do have the requisite sort of control over our beliefs, thereby rejecting (2); ii) they can argue that deontological judgments do not have voluntarist implications, thereby rejecting premise (1); or iii) they can accept the argument and admit that the familiar deontological terms of epistemic appraisal really are inapplicable." This in itself is a surprising conclusion, whether or not a deontological analysis of justification is accept- able.

"' Matthias Steup, "The of Epistemic Justification," Chapter 4 of Contenzporctry E11istemology (Upper Saddle River: Prentice-Hall, 1996). I' This last alternativc implies that either epistemic justification is not to be analyzed in deontological terms or else that epistemic justification is itself inapplicable to our beliefs, just like the other deontological epistemic terms.

SYMPOSIUM 669 I will discuss response (i) in section B below and response (ii) in section C. I will not discuss response (iii) except in passing and by implication.

B. Doxastic Voluntarism In this section I will examine some arguments for doxastic voluntarism and some arguments against it. I will eventually defend a version of doxastic voluntarism and I will also argue that this is of absolutely no epistemo- logical signiticance and that it does to help resolve the voluntarism puzzle. Although some have argued that it is a conceptual impossibility that anyone form a belief voluntarily, I will not discuss that view.I2 Instead, I will examine the claim that it is a contingent matter of fact that we lack that ability.

BI. Tlze Argunzent for Involuntarisnz Alston has given the most thorough defense of the contingent inability thesis, the thesis that as a contingent matter of fact people are not able to acquire beliefs voluntarily. Alston's paper includes an excellent survey of a variety of notions of voluntary control. For each type, he argues that we lack that sort of control over beliefs. (Alston does admit that there is one very weak of control that does apply to belief. But he contends that this sort of control does not provide an adequate basis for a good response to the Voluntarism Argument.) Alston begins by discussing basic voluntary control.'Ve have basic voluntary control over those actions that we can "just do." Simple bodily motions are the grime examples. I can just raise my hand, close my eyes, and bend my knee. Some people, but not I, can wriggle their ears and curl their tongue. Alston correctly says that forming a belief is not like that. We can't just do it at will. Alston turns next to nonbasic inznzediate voluntary control.14 One has this sort of control over the things one can do right away by doing something else, typically something over which one has basic voluntary control. Stan- dard examples are opening doors and turning on lights. We can, in typical circumstances, do these things simply by moving our bodies in the appropri- ate ways. There's here concerning what counts as "right away" but l2 For defenses of the conceptual impossibility claim, see Bernard Williams, "Deciding to Believe," of the '(New York: Cambridge University Press, 1973), Jonathan Bennett, "Why Is Belief Involuntary'?," Ancrlysis 50 (1990) and Dion Scott-Kakures, "On Belief and the Captivity of the Will," Philosol~hycrnd Pherzon~etzologicalResearch 53 (1993). For discussion of Williams, see Barbara Winters, "Believing at Will," Jourtzltrl of Philoso11lzy 76 (1979). For a discussion of Scott-Kakures' article, see Dana Radcliffe, "Scott-Kakures on Believing at Will," Philosophy urzd Phenometzolo;i.iccrl Resectrch 57, 1 (1997). l3 "The Deontological Conception of Epistemic Justification," Section 111, pp. 263-68. l4 See Section IV, pp. 268-74.

670 RICHARD FELDMAN that vagueness is in no way problematic. This is because the boundary between nonbasic immediate voluntary control and the next weaker kind of control, long range voluntary control, is acceptably imprecise.15 The sorts of things over which we have long range voluntary control are the sorts of things we can do over time by doing other things. Perhaps my house is an example. Or, more precisely, I have long range voluntary control over what color my house is because I can do things like paint it. Finally, there is indirect voluntary injuence.'"his is the kind of control we have when we can undertake a course of action that may some condition over the long term. Perhaps a person has indirect voluntary influence over the condition of her heart, since diet and exercise (courses of action we can more directly control) can affect it. Alston claims that we have only this weakest sort of control over our beliefs. One can engage in a course of action for the purpose of inculcating or losing a particular belief. Consider my belief that the Earth is not flat. There's nothing I can do to get myself to stop believing that right away. It's not like shutting the door or turning out the lights. And there's nothing much I can do long range to control it either. I might enroll in the Flat Earth Society, read conspiracy literature asserting that satellite photos are all phony, and so on. Perhaps this will help rid me of my belief. Alston would agree that it might. But this gets us, at most, indirect voluntary influence, and this is not the sort of el'fective voluntary control required to refute The Voluntarism Argument. l7 I believe that, for most part, what Alston says is right. However, I will argue that we have considerably more control over some of our beliefs than Alston acknowledges. Still, this control does not undermine the basic idea behind The Voluntarism Argument, although it may show that the argument reformulation.

82. An Argument for Doxastic Voluntarisnz My argument for doxastic voluntarism begins with the assumption that there are states of over which people have nonbasic voluntary control. For example, I have nonbasic voluntary control over whether the lights in my office are on. All I have to do is move in a certain way to get the lights on or off. And I can do this. The next step of the argument notes that my belief about whether the lights are on tracks their actual almost perfectly. As a result, I have a similar amount of control over whether I believe that the lights are on. All I have to do is move a certain way and then

Is See Scction V, pp. 274-77. '' See Section VI, pp. 277-83. " Perhaps this point suggests that both (2) and the consequent of (1) should be about eJfeclive voluntary control.

SYMPOSIUM 67 1 I'll have the relevant belief. More generally, when I have control over a state of the world and my beliefs about that state track that state, then I have just as much control over my belief about the state as I have over the state itself. Thus, we have nonbasic immediate voluntary control over our beliefs about states of the world over which we have control, provided our beliefs are responsive to those states. Furthermore, if we know that we will respond in some mistaken way to some state of the world over which we have control, we also have control over the resulting (erroneous) belief. So, doxastic volun- tarism is true after all. Premise (2) of The Voluntarism Argument is false. I believe that the of nonbasic voluntary control over beliefs can have prudential and moral significance. If the department chair announces that she'll give a raise to all and only those members of the department who in 30 seconds believe that the lights in their office are on, I'll head for the light switch and turn on the lights to make sure I have the desired belief. If the chair perversely announces that the graduate students will be mercilessly tortured-say, by forced to take additional prelims-unless in 30 seconds I believe that my lights are on, then I'd better make sure that I have that belief. I'm in control. Thus, we do have control over many of our beliefs. Of course, we don't often exercise this control. That is, we don't often believe things voluntarily. And this leaves this defense of doxastic voluntarism without a great deal epis- temological significance. The central for this is that we make deonto- logical epistemic judgments about beliefs that we can't control and these judgments are as routine and commonplace as are judgments about beliefs that we can control. Thus, our ability to control what we believe in the way described here is epistemically insignificant. We can take this into account by reformulating The Voluntarism Argument:

The Voluntarisnz Argument (Revised)

1. If deontological judgments about beliefs concerning states of the world people can't control are true, then people have (effective) voluntary control over those beliefs.

2. People do not have (effective) voluntary control over those beliefs

3. Deontological judgments about those beliefs concerning are not true.

This argument is as troubling as the original. It implies that an enormous of the deontological epistemic judgments we routinely make cannot be true. Furthermore, the argument might be extended to beliefs that we can control but have not formed in this voluntary way. Thus, I don't think that

672 RICHARD FELDMAN the fact that we have nonbasic immediate voluntary control over some beliefs provides the basis for an effective defense of epistemic deontologi~m.'~

B3. A More Robust From of Doxastic Voluntarism Compatibilists in the debate contend that we voluntarily perform an action when the action has the right sort of cause. A defender of doxastic voluntarism can argue that, analogously, we believe voluntarily when we believe as a result of the right sort of causal process. Roughly, the idea is that when unconstrained deliberation about leads to belief, we believe voluntarily." This is an interesting line of , and it is difficult to refute because we lack a fully adequate understanding of what counts as "the right sort of causal process." Still, I think that there are good grounds to reject the idea that the deliberative process that typically leads to belief is the kind of process that makes its product a voluntary action. A key fact about the clearest cases of voluntary action is that the action is caused by an intention to perform that action. I turn on the lights because I intend to turn on the lights; I type a certain word because I intend to type that word. No doubt there are puzzles about the details of these matters, such as puzzles resulting from the fact that our intentions are not as precise as the resulting actions-I not have intended to turn on the light in exactly the way I did. But however wc deal with thcsc details, the case of bclicf is dramat- ically different. We simply don't, in the typical case, form a belief as a result of an intention to form that belief. It may be that in some sense we do control what we believe, in much the way we control other involuntary processes in us. However, epistemic deliberation does not result in effective intentions to believe. Except in rare cases, we don't form intentions to believe. But such intentions are essential to voluntary control.20The cases of voluntary belief formation that I described earlier make it easy to see what it really would be like to believe voluntarily. In those cases, I do believe as a result of an intention to believe. But those are, for this very reason, quite unlike ordinary cases. Furthermore, if this defense of doxastic voluntarism were sound, then it seems that many other plainly involuntary behaviors would also be volun- tary. Deliberating about something can result in states other than belief. It l8 If this revised argument is strong, then it would also be reasonable to reject deontological analyses of epistemic justification as well. " For a defense of this view, see Matthias Steup, "Doxastic Voluntarism and Epistemic Deontology," forthcoming. For a similar view, see James Montmarquet, "The Voluntariness of Belief," Analysis 46 (1986). Bruce Russell endorses Steup's view in "Epistemic and Moral Duty," forthcoming. '" Perhaps the process in which by which we formulate intentions and act on their basis can become automated in certain ways, so that voluntary actions need not involve explicit conscious formulations of intentions. Belief formation is not like that either. It does not typically involve the formation of intentions to believe at all.

SYMPOSIUM 673 can result in desires or in panic. If we say that belief is voluntary because it is the outcome of deliberation about evidence, then it is hard to see why we shouldn't say that these other outcomes of deliberation are also voluntary. Yet it is clear that they are not. Again, what seems clearly missing is the right sort of intention. Thus, I reject this more robust form of doxastic voluntarism. There is good reason to say that the deterministic processes that typically lead to action render those actions voluntary. The processes lead to effective inten- tions to act. Deliberating about evidence does lead to belief, but not via such intentions. Such deliberation does not make belief voluntary. We need a better response to The Voluntarism Argument.

C. Ought and Can The second main sort of response to The Voluntarism Argument denies that the deontological judgments about beliefs imply that those beliefs are volun- tarily adopted. It denies premise (1) of the argument. It may be that there are differences among the various assertions that I've described as "deontolog- ical." I'll focus first on judgments about what one is obligated to believe and what one ought to believe and then discuss other judgments later in the section. One way to defend deontological epistemic judgments in the light of doxastic involuntarism is to argue that we can have epistemic obligations even though we can't fulfill them (or even if we can't help but fulfill them). This is to deny that "epistemically obligated to" or "epistemically ought to" implies "can". And one way to make this denial plausible to show that there are other kinds of ought statements that don't imply voluntary control. I will consider several candidates for this other kind of ought.

Cl. Contractual Obligations You can have an obligation to pay your mortgage even if you don't have the money to do so. Perhaps students in a class have an obligation to do the course work even if they are incapable of doing it. Other examples of this sort are rather easy to . Perhaps epistemic obligations are analogous to these financial and academic obligations. Though I once defended this view,21it now strikes me as an implausible model for epistemic obligations. The obligation to pay one's mortgage and the obligation to do one's course work are contractual obligations, although in the latter case the contract is in some sense implicit. It's difficult to see any basis for saying that we all have some sort of contractual obligation to believe things. Surely no such contract is explicit, and nothing analogous to enrolling in a course establishes an implicit contract.

- ~- 'I "Episternic Obligations," Philosophical Perspectives 2 (1988), pp. 24043

674 RICHARD FELDMAN C2. Obligations In a recent paper, says that there are two kinds of obli- gations, paradigm obligations and responsibility obligation^.^^ Only obliga- tions of the latter sort are associated with voluntariness. As examples of paradigm obligations, he presents:

1. "You ought to be walking on it in two weeksu-said by a physician as he finishes binding up a person's sprained ankle.

2. "That's strange; you ought to be seeing doublen- said by a psychol- ogist to his subject while conducting an in .

Wolterstorff suggests that epistemic obligations are similar to the obligations described by these . They lack any implication of voluntary control. No friend of epistemic deontologism should be comforted by the idea that epistemic obligations are like the obligations described by (1)-(2). This is because there are no obligations described by (1)-(2). Sentences (1)-(2) are "ought" sentences; they are not obligation sentences and they cannot be para- phrased in any straightforward way into obligation sentences. Your ankle has no obligation of any sort to heal; you have no "perceptual obligation" to see double. So, if there are epistemic obligations, they are not like the obliga- tions described here, since there are no obligations described here. Furthermore, the ought sentences Wolterstorff describes are not relevantly like epistemic oughts. (1) and (2) describe normal, or paradigmatic, behavior. Thus, for example, baring unforeseen developments, your ankle will heal in 2 weeks. But I don't think that this carries over to the epistemic case. That is, epistemic oughts don't describe paradigmatic or normal function. Some researchers report that people typically make various unjustified and predictably form unreasonable or erroneous beliefs.23Whatever the proper interpretation of this research actually is, it is at least possible that people normally make epistemic errors. It may be that we epistemically ought not do what we normally do. Epistemic oughts are not paradigm

22 Nicholas Wolterstorff, "Obligations of Belief: Two Concepts," in The Philosophy of Roderick Chisholm, Hahn, Lewis E (LaSalle, Illinois: Open Court, 1997). 2' For a discussion of many of the examples allegedly showing that people are irrational and an examination of their philosophical implications, see Edward Stein, Without Good Reason: The Debate in Philosophy and Cognitive (Oxford: Clarendon Press. 1996). 24 Thus, a researcher might say, while awaiting the subject's reply to a question that nearly everyone misses, "He ought to make the wrong here." The use of "ought" here is Wolterstorff's paradigm ought. But to describe the inference as "wrong" is to say that it is not the case that the person epistemically ought to do what the researcher expects hirn to do.

SYMPOSIUM 675 C3. Oughts There are oughts that result from one's playing a certain role or having a certain position. Teachers ought to explain things clearly. Parents ought to take care of their kids. Cyclists ought to move in various ways. Incompetent teachers, incapable parents, and untrained cyclists may be unable to do what they ought to do. Similarly, I'd say, forming beliefs is something people do. That is, we form beliefs in response to our in the world. Anyone engaged in this activity ought to do it right. In my view, what they ought to do is to follow their evidence (rather than wishes or fears). I suggest that epis- temic oughts are of this sort-they describe the right way to a certain role. Unlike Wolterstorff's paradigm oughts, these oughts are not based on what's normal or expected. They are based on what's good performance. Furthermore, it is plausible to say that the role of a believer is not one that we have any real choice about taking on. It is our plight to be believers. We ought to do it right. It doesn't matter that in some cases we are unable to do so. Thus, I reject the first premise of the revised Voluntarism Argument. Even in cases in which a believer has no control at all, it makes sense to speak of what he ought to believe and ought not believe. I think that most, possibly all, of the deontological terms we ordinarily use to evaluate beliefs can be explained in similar terms, although the case for some may be weaker than the case for others. Thus, if the standards of good believing allow either of two attitudes in a given situation, then it will be true that the believer has a right to either of those attitudes and that either of them is permitted.25 Other attitudes may be prohibited. It seems to me reasonable to say that when only one is permitted, then one has an episternic obligation to have that attitude. (I will add a qualification to this claim later in the paper.) It may be that some terms, especially those associ- ated with praise and blame, are to be reserved for voluntary behavior. Even here, the case is less than perfectly clear, since we do praise and blame people for attributes, such as , that they are unable to control. Thus, I conclude that deontological epistemic judgments can be true even if doxastic voluntarism is false.

C4. Is This Deontology? I want to contrast the view I'm defending from a different view about epis- temic deontologism. Possibly some critics primarily to the idea that to be epistemically justified in a belief is to do the best that one can do with respect to that belief. Their target really does imply voluntarism. I haven't attempted to defend that view, and I don't think that it is true. I've taken on the much more modest task of defending the legitimacy of the widespread use of deontological language about belief. My contention is that we can have

25 I'll return to sorne related issues in section 11, part A.

676 RICHARD FELDMAN epistemic requirements, permissions, and the like even if voluntarism is false. In his influential paper on this topic, Alston begins by saying that it is "natural" to understand epistemological terms in a "'deontological' way, as having to do with obligation, permission, requirement, blame, and the like."2"s an example of the view he wants to question, he mentions Carl Ginet's view that explains justification in terms of whether one is as confident as one ought to be.27But by the end of his paper Alston seems to be directing his attack more narrowly. In his concluding paragraph he charac- terizes epistemic deontologism as the view that analyzes "epistemic justification in terms of from blame for belie~ing."~~I haven't attempted to defend this sort of deontologism here. Moreover, I don't think that this more narrowly defined sort of deontologism is so natural or common. My defense is only of the possible truth of the deontological epis- temic claims with which Alston began his paper, claims to the effect that people can have epistemic rights, duties, permissions, etc. They can, no matter what the truth about doxastic voluntarism is.29 Given that deontological epistemic judgments can be true, I turn next to a vicw about the conditions under which they are true.

11. Epistemic Oughts A. Evidentialism In this section I will defend a variation on William K. Clifford's frequently quoted claim that "It is wrong, always, everywhere, and for anyone to believe anything upon insufficient eviden~e."~~'It is not my purpose here to explain or defend the thesis Clifford asserted with these words. My view may differ from Clifford's in any or all of the following ways. First, he may have been making a ~noralevaluation of believing on insufficient e~idence.~'The claim

2h "The Deontological Conception of Episternic Justification," p. 257. 27 Ibid., p. 259. 2X Ibid., p. 294. 2' It is possible that some critics will contend that sentences like "Smith ought to believe that the earth is not Rat" are ambiguous between deontological and non-deontological . They might then object that I've merely argued for the possible truth of the non- deontological interpretation of the sentence. I don't see any reason to admit that there are multiple senses of this sort. The deontological sense of the sentence just mentioned seems to be nothing other than the conjunction of the non-deontological sense with the proposition that it is up to Smith whether Smith does believe the earth is not Rat. Obviously, so interpreted, it does imply voluntarism. But the question with which we began was whether the original clairn was compatible with voluntarism. I don't think that this co~npatibilitycan be ruled out by fiat. W. K. Clifford, "The Ethics of Belief," reprinted in The of 2nd Edition, edited by Louis P. Pojman, (Belmont, California: Wadsworth, 1999), pp. 551-54. " In "The Ethics of Belief Reconsidered" Susan Haack argues that Clifford did not distinguish epistemic and moral senses of the key terms.

SYMPOSIUM 677 I want to make concerns epistemic rather than moral . Second, Clifford evaluates believing on insufficient evidence as wrong whereas the claim I will defend is that one ought not believe on insufficient evidence. Whether there is a between a thing being wrong and being some- thing one ought not do is unclear to me. If there is a difference, it is the latter claim that I want to defend. Third, in the quoted passage Clifford only objects to believing on insufficient evidence. He does not say that one ought to believe when one does have sufficient evidence. However, I do want to make this additional claim. A succinct way of stating my thesis is that one always ought to follow one's evidence. This evidentialist thesis is what I will defend.32 The following principle provides a preliminary statement of the evidential- ist idea:

01. For any proposition p, time t, and person S, S epistemically ought to have at t the attitude toward p that is supported by S's evidence at t.

The attitude in question can be belief, disbelief, or .33 It's important to realize that (01) does not imply that a person ought to believe a proposition simply because some of the person's evidence supports that proposition. Belief is the attitude to have only if the evidence on balance supports p. Some philosophers who are generally sympathetic to evidentialism may think that merely having one's evidence on balance support a proposition is insufficient for it to be true that one epistemically ought to believe the proposition. That evidential support might be exceedingly weak, and some will think that stronger evidence is required. Those who insist on such a condition can interpret (01) in a way that fits with this idea. On this interpre- tation, believing p is supported by one's evidence only if the evidence supports p sufficiently well. There is, however, reason to doubt (01). Suppose that a person has evidence that conclusively establishes some proposition, q. There are then a huge number, perhaps an infinite number, of obvious logical consequences of q that are also supported by this evidence. For example, the evidence supports q disjoined with every other proposition, or at least every other proposition S understands. (01) seems to imply that S ought to believe all these disjunc- tions. Many of these disjunctions will be trivial; many will be of no practical

32 In "Evidentialism," Philosophical Studies 48 (1985): 15-34, Earl Conee and I defend the evidentialist thesis that believing a proposition is justified if and only if believing the proposition "fits" one's evidence. 33 Those who think that beliefs come in degrees may wish to replace this principle by one saying that one ought to believe to the degree that they are supported by their evidence.

678 RICHARD FELDMAN or theoretical interest. It may be that in some sense believing these proposi- tions is justified-the person has good support for them-but it is not true that he ought to believe them. To do so would constitute an exceedingly foolish use of S's cognitive resource^.^^ Whether this objection is a good one depends in part upon what exactly goes into believing a proposition. The more dispositional belief is, the less that (01) demands of us. If the person with conclusive evidence for q can satisfy its demands by being disposed to assent to each of the disjunctions if asked about them, then (01) doesn't really require that he expend any cogni- tive resources. He doesn't have to do anything to meet its standards. If, on the other hand, it takes more than that to believe a proposition, then perhaps (01) is excessive. I'm inclined to think that (01) does ask too much of us, but I won't argue for that here. Instead, I propose the following refinement that avoids the problem:

02. For any person S, time t, and proposition p, if S has any doxastic attitude at all toward p at t and S's evidence at t supports p, then S epistemically ought to have the attitude toward p supported by S's evidence at t.

(02) in effect conjoins three : if a person is going to adopt any attitude toward a proposition, then that person ought to believe it if his current evidence supports it, disbelieve it if his current evidence is against it, and suspend judgment about it if his evidence is ncutral (or close to neutral). A person might ncver even consider a proposition and not be about to adopt any attitude toward it. In that case, (02) docs not imply that there is any attitude the person ought to have toward the proposition. Thus, (02) avoids the problem concerning thc use of cognitivc resources that (01) faced. (02) merely states which attitude to take toward propositions about which you are going to have one attitude or another. It does not instruct you to believe logical consequences that you don't entertain or to otherwise squander precious resources.

B. Requirements and Perinissions (02) helps clear up a confusing issue about whether epistemic principles state what you are required (ought) to believe or what you are permitted to believe. Suppose some unimportant and unconsidered proposition obviously follows from your evidence. For the just stated, this fact conjoined with (02) doesn't imply that you ought to believe it. But (02) does imply that if you take any attitude toward the proposition, it ought to be belief. So, it's not

34 Alvin Goldman makes this point in "Epistemics: The Regulative Theory of ," The Journc~lof Philosophy 75 (1978): 509-23. just that believing it is permitted. That's the only pcrmissible attitudc. It's the one you ought to have if you have any attitude. Thus, no attitude is epistemically required, but only one is epistemically permitted. (02) thus rules out the possibility that each of two attitudes toward a proposition is permitted at a time. Some critics might regard this as a mistake. To see whether it is, let's begin by considering a moral analogue of this possibility. On standard views, there are cases of tics in : each of two actions is morally pcrmissible and neither is obligatory. It's pcrmissible for me to give my extra money to the Cancer Society and it's permissible to give it to the Heart Association. It's not permissible to spend it frivolously. It follows then, that it is permissible to give my moncy to the Heart Associ- ation and also pcrmissible to refrain from giving my moncy to the Heart Association. Each of two incompatible actions is thus permissible. This is a consequence of the existence of moral ties. If the epistemic case is similar, then there arc situations in which cach of two attitudes toward one proposi- tion is permissible. But I think that in the epistemic world, things aren't quite the same. There are no tics, in the relevant sense. Potential cases of epistemic ties arc of three sorts. The first, and easiest to dispose of, are cases in which the evidence for and against a proposition is equally weighty. In that case, evidentialism implies that the sole acceptable attitude is suspending judgment. Neither belief nor disbelief is permitted. That strikes me as the right result. The second kind of case consists in those in which two competing propo- sitions are equally well supported by the evidence. For example, there is equally good evidence supporting the of each of two suspects for a crime that could have been committed by only one person. Casual reflection might suggest that you could believe that either is guilty, but obviously, not both. In that case, it is permissible to believe that suspect A is guilty and suspect B is not, and it is permissible to believe that suspect B is guilty and A is not. In fact, however, I think that in the situation described you should not believe that either suspect is guilty. Again, if you have an attitudc toward these propositions, you should suspend judgment about cach, though perhaps not about their disjunction. There are possible cases like the one just described in which some action is forced on you and it seems that you ought to form a belief about which action to take. Suppose that you know that one of two boxes in front of you has a bomb that is about to explode. You havc time to opcn only one box and disarm the bomb if it's there. You have no reason to prefer one box to the other. Surely you ought not suspend judgment about the location of the bomb and do nothing. That evaluation is corrcct, but it's corrcct because you ought not do nothing. Failing to act would be imprudent, perhaps immoral. But you needn't havc a belief about the location of the bomb in order to act. You can simply choose to opcn one box. And if you arc the sort of person

680 RICHARD FELDMAN who can only pick a box in this case if you havc the relevant beliek thcn there may well be a scnsc in which you ought to form a belicf about where the bomb is (if you can). But the ought herc is a prudential or moral ought, not an cpistcmic YOUepistemically ought to suspend judgmcnt. The third and most difficult sort of casc to think about is one in which you have somc modcst amount of evidence supporting a proposition. It's tempting to think that it's pcrmissible to belicve the proposition and it's also pcrrnissible to suspcnd judgment about it, on the grounds that belicf requires morc than just this modest amount of evidence. In support of this linc of thought is the idca that bclicving on only modcst amounts of cvidence involves taking somc cpistcmic risk. Thcre is no unique amount of risk that is right or rational. Rather, people simply havc varying attitudes toward this risk. My own inclination, which I won't dcfcnd hcrc, is that you should bclicve when your evidence is supportivc rather than neutral, cven if the evidencc is not at all dccisivc. Those who say that onc's cvidence supports believing a proposition only when that evidcnce is sufficiently strong will in all bc faccd with bordcrline cases. In a case in which this modcst amount of evidence is slightly in favor of a proposition, they might think that believing it is permissiblc and that suspcnding judgment is permissiblc. Of coursc, they are not forccd to this conclusion, sincc thcy might hold that it is simply indeterminate which attitude one is permitted to havc and that it is not simply true that either attitudc is ~ermitted.~~ Thc cvidcntialist account of what we ought to bclicvc rclies crucially on thc notion of cvidential support. Analyzing this notion in a fully satisfactory way is no easy task. Among thc problems to be workcd out is that of determining which logical conscqucnccs of a body of evidence are supported by that cvidcnce. Thcrc arc possiblc cascs in which a person has evidencc that implies some proposition, but thc conncction between that evidence and that conscquencc is distant and difficult to see. It may bc wcll bcyond the talents of thc pcrson. I believe that in such cascs thc pcrson ought not belicve thc consequcncc. Given his failure to sec that it is a conscquence, to belicvc it (barring othcr reasons to belicve it) would bc rash. Furthermore, as I undcrstand (02), it has exactly the right rcsult in this sort of case. Thc fact that a pcrson's cvidcnce implies somc proposition is not sufficient for thc cvidencc to provide evidcntial support for the proposition. Roughly, only those propositions whose connection to the cvidence the person apprehcnds

" Perhaps the reason it seerns right to say that "You ought to forrn the belief if you ctrrz" is that the "ought" in this judgment is one that does imply "can". This is further evidence that the intended judgment here is not an episternic judgment. ' Hud Hudson helped me to see this point. Notice that philosophers who think that ties are possible can still be evidentialists. That is, they can say that you always follow your evidence and never believe contrary to your evidence. They just think that there are cases in which the evidence leaves open two possibilities and there's nothing else that could elirninate one.

SYMPOSIUM 681 are actually supported by his cvidcnce. And I think ascertaining this conncction is itself an element of the pcrson's evidencc. Thcse issues arc complex, howevcr, and I will not pursuc thcm furthcr hcre. This complctes my dcscription of thc evidentialist idca that one cpistemi- cally ought always to follow one's cvidence. I turn now to a dcfense of this idea.

C. Evidentialism and Epistemic One way to explain why wc ought to do something is to show that it is a means to some goal that wc havc. A similar, but slightly diffcrcnt, way is to show that it is a mcans to somc valuablc rcsult. Both alternatives cxplain what we ought to do instrumentally. To usc the former altcrnative to show that we ought to follow our evidence would rcquire showing that pcoplc have some goal-true belief or knowlcdgc, perhaps-and that following thcir cvidencc is thc propcr mcans toward that goal. If thc oughts in qucstion arc supposed to bc means to goals that peoplc actually havc, then it sccms that only pcoplc who do havc thc epistemic goals just mentioned would bc subjcct to the rclevant epistemic rcquircmcnts. However, (02) is not rcstricted in that way. It says that all pcoplc cpistcmically ought to follow their cvidencc, not just those who have adoptcd some spccifically epistcmic goals.17 Onc might argue that all peoplc naturally have epistemic goals or perhaps that the goal of true belicf or knowlcdge is in some scnsc an "" goal. I won't pursuc thosc claims here. Instead, I want to defend (02) by arguing that following one's cvidencc is thc proper way to achicvc something of cpistcmic value. My approach docs not depend upon thc assumption that epistcmic valuc is a kind of instrumcntal value. Of course, thc vicw I'll dcfend does not imply that beliefs having cpistcmic value lack any sort of instrumental valuc.

CI. True Belief and Epistemic Value Thc dominant sourcc in the literature for discussions of epistemic goals and epistemic value is Wi!liam Jamcs' famous passage:

There are two ways of looking at our duty in the matter of opinion,-ways entirely different, yet ways about whose difference the theory of knowledge seerns hitherto to have shown little concern. We riiust know the truth; and we nzust c~voiderror,-these are our first and great commandments as would-be knowers.. .'"

37 One could take epistemic oughts to be conditional. To say that one epistemically ought to believe something is to say that if one has episternic goals, then one ought to believe it. 3R , "The Will to Believe," reprinted in The Theory of Knowledge, 2nd Edition, pp. 55.5-62.

682 RICHARD FELDMAN One idca commonly cxtracted from this is that our epistemic goal is to believe all and only .3y But if this implies that all pcople actually have the goal of believing all truths, then I doubt that it is true. Pcople's goals are a varicd lot, and I doubt that all pcople have this one. Moreover, bclicving all truths is obviously an unattainable goal. We simply can't bclievc all the truths. Furthermore, attaining it is not dcsirablc. As anyonc who lived through President Clinton's impeachment trial knows, thcre are many truths one would prefer not believing (or even considering). So, it is doubtful that believing all truths is an actual, attainable, or desirable goal. Whatevcr our goals, it makcs scnse to suppose that bclieving truths (or somc rclatcd state) has cpistcmic valuc, that it is a good thing from an epis- temic point of view. I turn now to the idea that having true beliefs and avoid- ing falsc beliefs has epistemic value and we should follow our evidcnce because it is the best way to have valuablc belicfs. Considcr, then, the following principle:

V1. Each person maximizes epistemic value by making it the case that for every proposition p, he or she p if p is true and does not bclievc p if p is false.

Onc might thcn arguc that pcoplc can bcst maximizc cpistcmic value by following their evidence. There are a few problems with a dcfense of (02) that is bascd on (Vl). For onc thing, (Vl) would sccm bcttcr suitcd to a dcfcnsc of (Ol), since (Vl) assigns value to belicving all truths, not just to the ones about which one has somc attitudc or othcr. Furthcrmore, the allegcd connection between (VI) and (02) is bascd on the assumption that following your cvidcncc is the best way to get at thc truth. And I just don't see why this is true. If you are in unfortunatc circumstances in which the information you have will lead you to falsehoods, following your evidence is not the best way to the truth. That is, it is not thc most cffective way to get at thc truth In these circumstances, ignoring your evidence is a bctter way to believe (some) truths. A possible reply to at least the second problem just mentioned for the argumcnt from (Vl) to (02) rclies on the claim that while following one's evidence might not be an infallible way to achieve the goal stated in (Vl) or cvcn always thc most cffcctivc way, it is ncvcrthcless thc bcst or most reasonable way to try to havc truc beliefs. This defense of (02) is quitc simi- lar to the one I will prcsent in section (C3) below and I will discuss it there

3"he idea that this is our epistemic goal is widespread. It is the starting point for numerous discussions of a variety of epistemological issues. See, for example, Alston's "The Deontological Conception of Justification," p. 258 and Richard Foley's The Theory of Episremic Kationuliry (Cambridge, Massachusetts: Harvard University Press, 1987),p. 8.

SYMPOSIUM 683 C2. Knowledge and Episternic Value What James actually said in his famous passagc wasn't that you should bclicvc truths or avoid belicving falsehoods. Nor did he say that you should havc truc bclicf or thc avoidance of false belief as a goal. What he actually said is that our cornmandmcnt, as would bc knowers, is to know thc truth and to avoid crror. Onc might cxtract from this the idea that what has cpistemic value is not mere true belief but rather knowledgc. Hcrc's onc way to spell out this idca:

V2. Each pcrson maximizes cpistcmic value by making it the casc that for cvcry proposition p, he or she knows p if p is true and knows -p if p is false.

(V2), likc (Vl), sccms bcttcr suitcd to support somcthing likc (01) than (02), but I will ignorc this point for now. Onc argumcnt from (V2) to (02) goes as follows. Givcn that onc can havc knowlcdgc only if onc has good cvidcncc for what one bclicves, one can get beliefs with epistcmic valuc only by following one's evidence. In other words, following one's evidcncc is a ncccssary condition for getting what (V2) says has epistemic value. If wc cpistcmically ought to do whatever is necessary to obtain this cpistcmic valuc, thcn wc cpistcmically ought to follow our evidcncc. The idea that knowledge is what has epistemic value has a certa~nplausi- bility. Knowledge is valuable, and it makes sense to think that knowledge is a particularly episternic kind of valuc. Howcvcr, thc argument from (V2) to (02) is unsound. Thc ccntral problcm with thc argumcnt turns on cascs in which a person has strong cvidcncc for a falsc proposition, f. According to the evidentialist position cxprcsscd in (02), such a pcrson should believe that falsehood. And the argument for (02) based on (V2) relies on the assumption that the person should adopt the attitudc that is ncccssary for him to have in ordcr to havc knowlcdgc. But since f is false, there's an equally good argumcnt for the conclusion that the person ought to believe -f, since knowl- edge also rcquircs truth. If adopting an attitude that will yield knowledge is what's valuable, thcn in such a situation adopting onc attitudc-bclicving f- will satisfy onc ncccssary condition for knowledge and adopting a diffcrcnt attitude-bclicving -f-will satisfy a different necessary condition for knowl- cdgc. (V2) provides no rcason to think that onc ought to bclicvc f in this case. Thus, it fails to support (02). A rclatcd point concerns cascs in which onc has strong cvidcncc for a proposition, but cvidence insufficiently strong to yicld knowlcdgc. (02) implics that one ought to believ~in such a casc. But it can't bc that onc ought to bclicvc in such a case becausc that is a mcans to knowlcdgc. Thc situation precludes having knowlcdgc, yct (02) has implications rcgarding what onc ought to do. Again, (V2) fails to providc support for (02).

684 RICHARD FELDMAN C3. Reasonable Belief and Epistemic Value While true belicfs may havc considerable instrumental valuc, a person who irrationally believes a lot of truths is not doing well epistemically. In contrast, a person who forms a lot of rational but false beliefs is doing well epistemically. While knowledge also has a kind of value, seeing it as the only thing of epistemic value fails to explain what is valuable about forming beliefs that fall short of knowledge. We avoid the problems associated with identifying epistemic value with true belief or with knowledge if instead wc say that what has epistemic value are rational beliefs. To do well as a believer, to achieve a kind of epistemic excellence, one must form only rational beliefs. I will discuss a defense of (02) based on this account of epistemic value in this scction. One way to understand this evidentialist perspective on epistemic value and epistemic oughts is as follows. Consider a person who is contemplating a particular proposition. To carry out the role of being a believer in an cpistemically good way, in a way that maximizes epistemic value, the person must adopt a rational attitude toward this proposition. There are other values that beliefs might yicld. Some beliefs might havc prudential or moral value. They might make people feel good or provide comfort for others. They might help one to undertake risky but beneficial behavior. They might give one self-confidence that can be help to advance one's career. But beliefs that are beneficial in these ways can nevertheless fail to be rational. They can lack epistemic value. To achieve epistemic valuc one must, in cach case, follow one's evidence. Here, then, is a principle about epistemic value that supports (02):

V3. When adopting (or maintaining) an attitude toward a proposition, p, a person maximizes epistemic value by adopting (or maintaining) a rational attitude toward p.

Given that (V3) specifies what max~mizesepistemic value, a believer epistemically ought to form attitudes as directed by (02): he ought to follow his evidence. This defense of (02) depends on the substantive epistemological thesis that rationality consists in making one's beliefs conform to one's evidence. This thesis is apparcntly denied by some, for example some reliabilists. If has substantive implications that differ from those of evidentialism, it must be because in some cases it implies that it is rational to form beliefs in the absencc of evidential support. I will not under- take a defense of this evidentialist claim here, nor will I discuss the reliabilist alternative. Anyone who is about to adopt an attitude toward a proposition, p, and who adheres to the dictates of (02) will get knowlcdge (of p or of -p) when- ever such knowledge is attainable. If the available evidence is strong enough

SYMPOSIUM 685 to yield a belief well-enough justified for knowlcdge, and the other conditions for knowledgc are satisfied, then the person will have knowlcdge. Wherc knowledgc is not attainable, thc pcrson will havc reasonablc belief. He'll be doing as well as hc can, cpistcmically spcaking. By sccking rational belicf, thcn, one will get knowledge wherc one can. And, of coursc, unlcss one is in unfortunatc circumstances in which one's cvidcnce frcqucntly lcads to false beliefs, anyone who follows (02) will mostly have true belicfs. (V3) and (02) concern thcmsclvcs with how to maximizc cpistemic value when adopting an attitudc toward a proposition. Thcre arc many ways in which onc might bchavc ovcr thc long term that will hclp one to gain knowledge of important . Thc relcvant bchavior might include enrolling in suitable courscs of study, maintaining a hcalthy lifc-stylc to kecp onc's sharp, and cultivating sound infcrcntial habits. Evidcntialism is silent about such practices. It focuses on the epistemic value to be obtaincd immediately from the adoption of an attitude toward a proposition. The way to do that, in evcry casc, is to follow the cvidcnce onc has. At thc cnd of section (Cl) I mentioned the idea that one might defend (02) on the grounds that having truc belicfs is epistcmically valuable and that following one's cvidencc is thc most reasonable way to try to gct bclicfs that have epistcmic valuc. Thc currcnt vicw is that rcasonablc bclicfs are cpistem- ically valuable and that following one's evidencc is a pcrfect mcans to getting valuable bcliefs. There arc then two ways to support (02). The dlffcrences bctwcen thcsc dcfcnses of (02) may not bc grcat, but thcre arc two points worthy of notc. Thcy can bc brought out by considcring a pcrson whose evidcncc supports a grcat many falsehoods. First, supposc the pcrson follows his evidencc. According to both vicws, this pcrson is bclicving as hc ought. But according to the earlier vicw his beliefs, being false, lack epistcmic valuc whercas according to the current view they are epistemically valuable. Sccond, suppose that the pcrson docs not follow his evidcnce. In that casc, both views imply that hc is not bclieving as hc ought, but thc earlicr view implics that hc is, by luck, achieving epistemically valuable beliefs. In both cascs, I find thc implications of currcnt vicw morc appealing. I don't scc any- thing epistemically good about the person who irrationally gets true bcliefs. I don't think that it would be correct to say of him that he's achieved epistemic excellencc, cvcn though hc's donc it in an irrational way or mcrcly by luck. Rather, I think hc's failcd cpistcmically, not only bccausc hc isn't bclicving as he ought but becausc he docs not havc rational bclicfs. Of coursc, therc may bc some instrumental valuc in those truc bcliefs. They may hclp thc person negotiatc the world in a better way. But that is a differcnt mattcr.

686 RICHARD FELDMAN D. Three Objections to Evidentialism

Dl. Evidence One SIzould Have Had

Considcr a person who is negligent about collecting cvidence. Suppose that I have thc firm belief that

G. Taking ginkgo supplements is a safc and effective way to improve my .

I have a modest amount of evidence supporting (G). I then see on the cover of a reputable magazine that it contains a major article about the merits of gingko. Rather than read the article, I avoid it for fear that it will undermine a belief I prefer to keep. Thus, I am negligent in collecting evidence; there is evidence I don't have but should have. Furthermore, suppose that if I read the article, I would have acquired strong evidence against (G). Thus, while (G) is supported by the evidence I do have, I should have had additional evidence and I shouldn't believe (G). So, (02) has the wrong result in this sort of casc. It implics that I ought to follow thc limitcd evidcncc I do havc rather than the largcr body of evidcncc I havc ncgligcntly a~oided.~' The details of examples such as this one are quite important. Therc are possible cxamples in which a person has some good reasons to believe some- thing, but knows that there are availablc strong considcrations to thc contrary although he chooses not to makc himself familiar with that evidencc. Suppose, that the magazine headline in our example is "Gingko Shown to be Ineffective." Upon seeing this headline, I immediately stop reading since I don't want evidence like that. I continue to believe (G) on the basis of my old reasons. Surely my conduct in this case is reprchensible. But just as surely, I have acquired evidence against (G) and my bclicf losses some considerablc support the moment I see the title of the article. It gives me good rcason to think that there are strong objections to (G), even if I'm not yct in a position to say in any dctail what thcy arc. That significantly alters the evidential status of thc proposition for mc. Givcn thc credibility of the source and the of the article title, most likely my overall evidence no longer supports my belief. I no longer ought to believe (G). Variations on the example are possible. The article might be in a publica- tion whosc rcliability is cntircly unknown to me or in a magazinc I know to be thoroughly disreputable. Other variations on the cxamplc concern the wording of the title. It could be somcthing neutral such as "Some New Information on Gingko." Again I ignorc thc articlc. Again, assume that it

4" Keith DeRose presented an objection along these lines when he cornrnented on a version of this paper presented at the Rutgers Epistemology Conference.

SYMPOSIUM 687 contains the same devastating objections. In this case, the mere of the existence of the article has much less evidential . Nevertheless, you might plausibly think that in these cases, or at least some of them, I am terribly negligent in ignoring the article. I shouldn't do that. Had I not done it, I wouldn't have maintained my belief in (G). So, I ought not have that belief. The statement of the objection seems to me to state the heart of the reply. The name and source of the message do not provide me with much, if any, reason to stop believing (G).41Until I've read the article, it would be bizarre for me to stop believing what's supported by my old evidence. I don't have any reason to. Suppose that instead of negligently ignoring the article, I'm busy and I simply it aside until I can give it proper . There's nothing negligent about this behavior. What should I believe in the mean- time? Should I think, "Well, somebody has written something about gingko, so I'd better stop believing that it is effective"? I think that the answer is obvious: I shouldn't change for that reason. And this fact doesn't changc whcthcr my motivcs for sctting aside the article are laudatory or rcprchcnsi- ble. In cvcry case, the right answcr dcpcnds upon what evidence I already havc, including thc evidcncc about thc possiblc cxistencc of these objections. No mattcr what the answcrs arc to questions about how I ought to conduct my inquiry, where I ought to look for cvidcncc, and so on, therc always remain the questions, "What should I believe in thc mcantimc?" "What should I believe until I have a chance (or the couragc) to look at that new evidence?" It's that natural and central question to which evidentialism provides a good answer.42 One might think that when one should look at additional evidence, one should always suspend judgment in the meantime. But that is a clcar mistake, as the examplc above illustrates. You should follow your current cvidcncc. If you do get more evidence, then you should follow the combincd cvidcncc you have at that time.

02. Duty to Gather Evidence Even if it is true that in the examples just discussed I am believing as I ought given the limited evidence I have, it is plausible to think that there are other epistemic requirements that I ought to fulfill. In particular, in some versions of these cases, I ought to obtain the additional evidence that is available to mc. Morc gcnerally, in a widc varicty of cascs, it might be thought, one

41 One rnight argue that we all know that rnagazines usually run articles like this only when they are negative. If that's the case, then seeing the seemingly neutral title does provide me with some evidence against (G). The case is then similar to the first version in which the title is explicitly negative. 42 Richard Foley takes a similar approach to epistemic questions. See especially Chapter I of Wor.kirlg Without n Net (Oxford: Oxford University Press, 1993).

688 RICHARD FELDMAN ought to gather additional evidence about the propositions one considers. It's not enough simply to follow the evidence one actually has. Thus, evidential- ism overlooks important epistemic oughts. In a recent paper, "The Epistemic Duty to Seek More Evidence," Richard J. Hall and Charles R. Johnson argue that you have an epistemic duty to seek more evidence about every proposi- tion about which you are not certain.43While many epistemologists would reject the demanding condition Hall and Johnson describe, it may seem obvious that in many cases you ought to seek evidence regarding proposi- tions about which you will form beliefs. However, the evidentialist view defended here apparently conflicts with this seemingly obvious truth. If the fundamental epistemic goal is just to have reasonable beliefs, then nothing about evidence gathering techniques or the like follows as a means to that This goal just has implications concerning what attitudes we ought to take given the evidence we have. By seeking out new evidence concerning some important proposition and then believing what that evidence supports, I don't do a better job of achiev- ing the goal of believing reasonably. I achieve that goal at any moment by believing what is then supported by my evidence. It's surely true that there are times when one would be best off finding new evidence. But this always turns on what options one has, what one cares about, and other non-epistemic factors. As I see it, these are prudential or moral matters, not strictly epistemic matters. A familiar distinction may help make the evidentialist account seem more attractive. This is the distinction between short-term and long-term goals and the related distinction between synchronic and diachronic rationality. The former concerns questions of rationality at a given moment while the latter concerns rationality over time. Evidentialism is best seen as a theory about synchronic rationality. It holds that the epistemically rational thing to do at any moment is to follow the evidence you have at that moment. It doesn't address questions of how to conduct inquiry over periods of time. Thus, it does not address questions about how to gather evidence, when one ought to seek additional evidence, and so on. In my view these diachronic questions are moral or prudential questions rather than epistemic questions. You should gather more evidence concerning a proposition only when having a true belief about the subject matter of the proposition makes a moral or prudential dif- ference and gathering more evidence is likely to improve your chances of

43 See Richard J. Hall and Charles R. Johnson, "The Epistemic Duty to Seek More Evidence," Arnericcltz PRilosopt~icalQilnrterly 35 (1998): 129-40. Hall and Johnson actually defend the thesis that this remarkable duty follows from the assumption that our goal is to believe all and only true propositions. They do not defend the claim that this is our goal. 44 Similarly, if the goal is just to now have true beliefs about the propositions about which we are about to form attitudes, then there's no need to gather new evidence.

SYMPOSIUM 689 getting it right. Of course, whether you ought to gather such evidence also depends upon what other things you could do with your time. Epistemolog- ical considerations simply don't resolve such matters. There are cases in which one can spend one's time gathering evidence about propositions concerning inconsequential and trivial propositions or about more weighty matters. Evidentialism provides no guidance about what to do. As I see it, this is not a weakness of evidentialism, since such choices are not to be made on epistemic grounds. What topics you ought to investi- gate depend upon what topics are of interest to you, what investigations can help you to make your own life or the lives of others better, and other such matters. Evidentialism is silent on those moral and prudential issues, and I don't see why it should address them.

D3. Being Kutional By Avoiding Evidence The evidentialist principles stated here imply that one can get oneself into a highly rational state by ridding oneself of as much evidence as one can and then suspending judgment about virtually that comes to mind. If a person finds a drug or a machine that can erase from his and arranges to be immersed in a sensory depravation tank, he'll have very little evidence regarding anything. By believing very little, he'll then be highly rational according to evidentialist standards. Yet, this may seem just the opposite of being rational.j5 Evidentialists are committed to the view that the person just described would have little evidence regarding anything and would be rational to suspend judgment about nearly everything. But that seems to be exactly the right conclusion. Once the person has lost his evidence, he has no reasons to believe much, and he'd be unreasonable if he did believe things that would have been well justified for him had he been in more normal circumstances. But he's not in normal circumstances and evidentialism concerns itself with assessing what he should believe in the circumstances he is in, not with what he should believe if he were in different circumstances. Critics might think that he shouldn't get himself in a condition of eviden- tial deprivation in the first place. Evidentialists can agree, though not on evidentialist grounds. By putting himself in those conditions, he's made him- self unable to act on behalf of others who may rely on him. Perhaps he morally ought to avoid rendering himself useless in this way. Furthermore, by putting himself in a condition of evidential deprivation, he deprives himself of all the pleasure that comes from experiencing the world. Unless his prior circumstances and prospects are extremely grim, it is likely to be imprudent to put himself in that situation. Evidentialists can agree with these critical of this behavior, but these are not epistemic evalua- j5 This objection was suggested to me by .

690 RICHARD FELDMAN tions. They are moral or prudential evaluations of behavior related to the formation of beliefs. I conclude that objections such as the ones considered here do not under- mine the evidentialist view of what we epistemically ought to do. Evidential- ism says that when adopting a doxastic attitude toward a proposition, a person epistemically ought to adopt that attitude that is supported by the evi- dence the person has at that time. How the person came to have that evidence, whether by conscientious inquiry or by avoiding potentially troublesome information, is irrelevant to this epistemic fact. Similarly, how the person ought to proceed in future inquiry is also irrelevant. Evaluations of this behavior can be made, of course, but these evaluations are of a different nature than those made on evidentialist grounds. The evidentialist evaluation assesses whether the attitude formed in the circumstances one is actually in is the attitude one ought to have formed in those circumstances.

111. Epistemology and Ethics There are several questions about the relation of epistemic duties (or oughts) to ethical and other duties. For example, there is in the literature considerable discussion of the idea that there's some important connection between what one is epistemically justified in believing and what one is morally justified in believing. On onc sidc, there is the line of thought suggested by Clifford, according to which believing without epistemic justification is always morally wrong. To defend this thesis, you'd have to argue that such believing is always voluntary or else that involuntary behavior can be morally wrong. I'm not inclined to argue for either of these theses. Other questions concern whether epistemic duties are a "special case" of moral duties and whether there can be cases in which there is a conflict between these different kinds of duties.4" won't take up these topics here. I will discuss one puzzling issue in this area. The question I do want to address can best be raised by attending to a remark Hall and Johnson make in defense of their claim that we have an epis- temic duty to seek more evidence for all uncertain propositions. They say that this duty "is probably a pretty weak duty; most moral and prudential duties would trump it in cases of conflict."47 They quote Chisholm as holding that moral duties always trump epistemic duties. What I want to discuss is whether any sense can be made of the claim that moral duties trump, or do not trump, epistemic duties. There is no problem with the idea that duties of the same kind can be weighed in importance. For example, a person might have a moral duty to

'"or '"or discussion of the latter question, see Eugene Mills, "The Unity of Justification," Philosophy nnd Plzenotizenologicc~/Reseccrch 58 (1998): 27-50. For discussion of the former question, see Susan Haack, "The Ethics of Belief Reconsidered." 47 "The Epistelnic Duty to Seek More Evidence," p. 136. his students to show up for class and a moral duty to his sick child to stay home and care for her. And one might say that one of these duties outweighs, or trumps, the other. The idea here, I take it, is that there is some scale of moral value and that fulfilling one duty contributes more to that value than does fulfilling the other. What is far less clear is that there is such a thing as just plain ought, as opposed to the various kinds of oughts philosophers have succeeded in distin- g~ishing.~~To understand the issue, it's important to distinguish two very different points. Suppose you accept the view that it is morally wrong to believe on the basis of insufficient evidence. (Set aside worries about volun- tarism.) I take it that at most you would think that the fact that a particular action is a case of believing on insufficient evidence counts towards its being wrong. As some would put it, it is prirnu facie wrong. This leaves open the possibility that this case of believing on insufficient evidence has other features that count in its favor, morally speaking. It might be a case of being supportive of one's or disregarding some excusable flaws in others. It might have much to be said for it. I have no trouble at all understanding the idea that we can in principle compare the negative moral value of failing to follow one's evidence to the positive moral value of the other features of the case. But here we are comparing two moral factors to reach an overall moral assessment. I take it that when people say things such as "Moral oughts trump epis- temic oughts" they are not saying that the moral weight of epistemic oughts is less than the moral weight of other moral considerations. I believe that what they are saying is that there is some sort of generic ought that somehow encompasses moral considerations, epistemic considerations, and perhaps others, and then weighs them against one another to come up with an overall assessment. This is not any particular kind of ought. It is just plain ought. I take Hall and Johnson to be suggesting that when you epistemically ought to gather more evidence and you morally ought to do something else, the moral ought "wins" and you just plain ought to do that other thing. It's this that I just don't understand. Of course, by this I mean to suggest that no one else understands it either. It makes no sense. I know of no way to establish that the notion of just plain ought makes no sense. But I can give some partial defense to my view by noting that it would be a mistake to assume that there must be such a thing as "just plain ought" simply because there are various kinds of oughts. There needn't be anything meaningful about their combination. To see why, consider an . Suppose a child has two dolls. One is short and squat. The other is tall and thin. The child asks you which doll is "bigger". You are faced with a

4X I learned this way of formulating this question from an unpublished paper by Owen McLeod.

692 RICHARD FELDMAN problem. One doll is taller. The other is wider and, let's assume, more volu- minous. The child could be asking which is taller, which is wider, or which is more voluminous. Each of those measures could be expressed by "bigger". We needn't assume that there's any meaningful measure that combines these three, yielding "bigger all things considered." Or, more precisely, we could arbitrarily combine the various measures in any way we like. But it would be a mistake to pick any such measure as what really counts as "overall bigger." It would make no sense to ask whether height trumps width or volume trumps height in assessing overall size. We've disambiguated the word "bigger". There's no putting the senses back together, at least no way that isn't simply arbitrary. There's no such thing as "just plain bigger." There are simply the various bigger-than . Consider another example. Suppose we are considering the wealth and strength of two people, A and B. A is stronger, B is wealthier. We ought not assume that there's some proper way of combining these two measures into one, which we might call "strealth". We can of course combine measures of strength and wealth in a limitless number of ways. It might be that some combination would in fact best suit our purposes. But the mere fact that there are two measures, wealth and strength, does not guarantee that there is any proper way to combine them to form strealth. The existence of numerical measures of wealth and strength ought not deceive us here. We could measure one's wealth in rubles and one's strength in terms of the number oP grams one can bench press. We could then add those together to get a proposed measure of strealth. But we also could add one's wealth in dollars to the number of pounds one can bench press. That will give a different number. And there are a limitless number of other such sums. Each will yield poten- tial strealth rankings. The different possibilities will have the effect of count- ing one measure more heavily than the other. If someone were to object to one of these strealth measures on the grounds that it conflicts with the fact that wealth trumps strength in measures of strealth, I think we should say that we have no idea what that person is talking about. We have no indepen- dent concept of strealth for which there are correct weightings of the two components. I think that the same is true of the attempt to compare various kinds of oughts to determine what one just plain ought to do. Suppose that one belief is prudent for me-it will maximize my well being or what I care about or what is important to me-but it is not a belief I epistemically ought to have since I lack evidence for it. We can understand the idea that forming an unjustified belief might be imprudent, since it might foster a bad habit. So, we can imagine weighing the short-term prudential gain against the long term prudential cost. But I can see no values to which we could be appealing when we ask whether the overall prudential benefit trumps the epistemic cost.

SYMPOSIUM 693 For each "ought" there is an associated value. We ought, in the relevant sense, to do the thing that maximizes that value, or perhaps something that does well enough in achieving that kind of value. For example we morally ought to do what maximizes, or produces enough, moral value. We pruden- tially ought to do what maximizes, or produces enough, prudential value. If there is such a thing as "just plain ought" then there is a value associated with it. The thing we just plain ought to do is the thing that comes out highest, or high enough, according to that measure. It's far from clear what that value would be-it isn't to be identified with any of the more determi- nate kinds of value and there seems to be no uniquely correct (or range of correct) ways to combine moral, practical, epistemic, and other values. We've disambiguated "ought" and we can't put the various senses back together again. There is no meaningful question about whether epistemic oughts "trump" or are trumped by other oughts. Stewart Cohen has presented the following objection to this view.4y Suppose a child comes to her parents with a problem. She is contemplating some action and she's figured out that it is the prudent thing to do but not the moral thing to do. In other words, she realizes that she prudentially ought to do it, but she morally ought not do it. She asks, "What ought I do?" She is clearly looking for a way to resolve the apparent conflict. The view I've defended requires the parents to say that there is no meaningful question here. More precisely, when she uses the word "ought" in her question, she is either using it in one of the disambiguated senses, in which case she already knows the answer to her question, or else she is using it in the sense of "just plain ought", in which case she's not asking a meaningful question. In the latter case, the question is like that of child who wants to know which doll is really bigger.50 It is difficult to resist the inclination to want to give the child advice about how to conduct herself. And there is advice that one might give. The question invites further reflection on whether the moral evaluation is really correct, given that the act is harmful to her. The question also invites further reflection on whether the prudential calculation is really correct, given that immoral behavior is also often imprudent. And in the particular case of a child asking a parent such a question, there is the matter of whether the child will have parent's approval whichever way she acts. There is, finally, the question of what the parent predicts he or she would do in similar circum- stances. Thus, there are several meaningful questions that one might ask in the sort of case Cohen envisions. However, as I see it, none of them amounts to the question of what one just plain ought to do.

4y In conversation. Another possibility is that she is using "ought" in yet another more determinate sense. This option seems irrelevant to the present example.

694 RICHARD FELDMAN IV. Conclusion I've argued in this paper that deontological judgments about beliefs do not imply doxastic voluntarism. I've also argued that we do have a form of voluntary control over a substantial number of beliefs, but that this fact is of no great significance for epistemology. Epistemic evaluations do not depend upon whether we have or exercise this voluntary control. I've also argued, or asserted, that what we epistemically ought to do is follow our evidence: when we have a doxastic attitude toward a proposition, it ought to be the attitude that is supported by the evidence we have. I haven't attempted to say any- thing here about what evidential support is or what it is to have something as evidence. Those are topics for another occasion. I have claimed that there is nothing more that a person epistemically ought to do than to follow her evidence in this way. In particular, activities such as gathering additional evidence for propositions about which one is uncertain are not among the things one epistemically ought to do, even if they are on other grounds highly desirable. Finally, I've expressed about the meaningfulness of questions about whether or not epistemological considerations are out- weighed by moral or prudential considerations in figuring out what we ought to do all things considered. The topics I've addressed by no means exhaust the issues that could properly bc raiscd under the heading "thc cthics of bclicf." But I hope to havc made some small contribution to our thinking about the specific topics I have addre~sed.~'

In working on this paper I've benefitted greatly from numerous discussions with Earl Conee, John Bennett, Stewart Cohen, Jonathan Vogel, and all the students in my Fall 1998 epistemology seminar. I'm also grateful to John Greco and Matthias Steup for helpful comments. Earlier versions of the paper have been presented at the University of Rochester and at Rutgers Univcrsity. I'm grateful to the audiences on both occasions and to Keith DeRose for his provocative comments on the latter occasion.

SYMPOSIUM 695 http://www.jstor.org

LINKED CITATIONS - Page 1 of 4 -

You have printed the following article: The Ethics of Belief Richard Feldman Philosophy and Phenomenological Research, Vol. 60, No. 3. (May, 2000), pp. 667-695. Stable URL: http://links.jstor.org/sici?sici=0031-8205%28200005%2960%3A3%3C667%3ATEOB%3E2.0.CO%3B2-H

This article references the following linked citations. If you are trying to access articles from an off-campus location, you may be required to first logon via your library web site to access JSTOR. Please visit your library's website or contact a librarian to learn about options for remote access to JSTOR.

[Footnotes]

2 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

3 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

4 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

5 Firth and the Ethics of Belief Roderick M. Chisholm Philosophy and Phenomenological Research, Vol. 51, No. 1. (Mar., 1991), pp. 119-128. Stable URL: http://links.jstor.org/sici?sici=0031-8205%28199103%2951%3A1%3C119%3AFATEOB%3E2.0.CO%3B2-%23

NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org

LINKED CITATIONS - Page 2 of 4 -

5 Justified Belief and Epistemically Responsible Action Hilary Kornblith , Vol. 92, No. 1. (Jan., 1983), pp. 33-48. Stable URL: http://links.jstor.org/sici?sici=0031-8108%28198301%2992%3A1%3C33%3AJBAERA%3E2.0.CO%3B2-A

6 Internalism Exposed Alvin I. Goldman The Journal of Philosophy, Vol. 96, No. 6. (Jun., 1999), pp. 271-293. Stable URL: http://links.jstor.org/sici?sici=0022-362X%28199906%2996%3A6%3C271%3AIE%3E2.0.CO%3B2-L

8 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

12 Why Is Belief Involuntary? Jonathan Bennett Analysis, Vol. 50, No. 2. (Mar., 1990), pp. 87-107. Stable URL: http://links.jstor.org/sici?sici=0003-2638%28199003%2950%3A2%3C87%3AWIBI%3E2.0.CO%3B2-1

12 Believing at Will Barbara Winters The Journal of Philosophy, Vol. 76, No. 5. (May, 1979), pp. 243-256. Stable URL: http://links.jstor.org/sici?sici=0022-362X%28197905%2976%3A5%3C243%3ABAW%3E2.0.CO%3B2-K

12 Scott-Kakures on Believing at Will Dana Radcliffe Philosophy and Phenomenological Research, Vol. 57, No. 1. (Mar., 1997), pp. 145-151. Stable URL: http://links.jstor.org/sici?sici=0031-8205%28199703%2957%3A1%3C145%3ASOBAW%3E2.0.CO%3B2-9

NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org

LINKED CITATIONS - Page 3 of 4 -

13 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

21 Epistemic Obligations Richard Feldman Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 235-256. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C235%3AEO%3E2.0.CO%3B2-B

26 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

27 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

28 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

34 Epistemics: The Regulative Theory of Cognition Alvin I. Goldman The Journal of Philosophy, Vol. 75, No. 10, Seventy-Fifth Annual Meeting of the American Philosophical Association, Eastern Division. (Oct., 1978), pp. 509-523. Stable URL: http://links.jstor.org/sici?sici=0022-362X%28197810%2975%3A10%3C509%3AETRTOC%3E2.0.CO%3B2-M

NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org

LINKED CITATIONS - Page 4 of 4 -

39 The Deontological Conception of Epistemic Justification William P. Alston Philosophical Perspectives, Vol. 2, Epistemology. (1988), pp. 257-299. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281988%292%3C257%3ATDCOEJ%3E2.0.CO%3B2-E

46 The Unity of Justification Eugene Mills Philosophy and Phenomenological Research, Vol. 58, No. 1. (Mar., 1998), pp. 27-50. Stable URL: http://links.jstor.org/sici?sici=0031-8205%28199803%2958%3A1%3C27%3ATUOJ%3E2.0.CO%3B2-Q

NOTE: The reference numbering from the original has been maintained in this citation list.